1
\$\begingroup\$

Terminology:


I have been experimenting with of the shelf temperature sensor nodes. The exact device in question is an Aeon Labs LLC, Aeontec Z-Wave MultiSensor 6.

I got the information that to reach the accuracy defined in the specs (+-1 Celsius) the sensor nodes have to be recalibrated after installation. The recalibration is done by using an offset value to even the sensor node temperature reading with a referent temperature reading.

I can reason the calibration of analog sensors where the voltage levels of the power can't be known at production time hence additional calibration is required once the power source is available.

This is an of the shelf sensor node intended to be used as a consumer electronic product. I haven't yet witnessed cheap dedicated thermometers ever needing calibration.

Questions:

  • Under what circumstances does a temperature sensor need calibration? (given that there are so many sensor nodes that don't offer the calibration mechanism)
  • Is the precision of +-1 Celsius a difficult requirement to achieve and can be severely reduced (3 Celsius difference between nodes placed next to each other) just by the fact that the sensor has been moved to a different environment (example. from loft to basement)?
\$\endgroup\$
  • \$\begingroup\$ Part 2 of your questions is confusing. Calibration is location specific. \$\endgroup\$ – Sean Houlihane May 20 '16 at 9:41
2
\$\begingroup\$

Calibration here is not related to the electrical performance of the device. As an example, the DS18B20 has a typical accuracy +/- 0.5C, from a 3-5V supply.

For a room thermometer, calibration refers to reflecting the room temperature to the value reported by the sensor, and is as much as a consumer feature than a technical necessity. There will be a small amount of self-heating in the node, there may be local heat sources in the room (say a TV, or a PSU), there may be air currents which lead to cooling, and a node placed higher up will most likely read higher than one placed low down. Many of these factors are not stable which makes the claimed accuracy rather unimportant.

Think of the calibration as applying an offset so the perceived optimum room temperature results in a similar reading from all sensors.

For example, there is a A/C thermostat next to me which reports temperature to a precision of 0.1C. Its exposed to sunlight through the window, and is next to a door. I guess it has maybe 0.5W of self-heating, except when the backlight is on. At different times, with the same room fabric temperature, the thermostat will record different temperatures - and that is not a reflection on the accuracy of the sensing device. Making the measurement accurate does not add value, since a person in the room also feels temperature differently as the conditions vary.

\$\endgroup\$
  • \$\begingroup\$ Many of these factors are not stable which makes the claimed accuracy rather unimportant. What do you mean by this? What would be the meaning of a temperature measurement without an accuracy? \$\endgroup\$ – TheMeaningfulEngineer May 20 '16 at 9:50
  • \$\begingroup\$ For example, there is a A/C thermostat next to me which reports temperature to a precision of 0.1C. Its exposed to sunlight through the window, and is next to a door. I guess it has maybe 0.5W of self-heating, except when the backlight is on. At different times, with the same room fabric temperature, the thermostat will record different temperatures - and that is not a reflection on the accuracy of the sensing device. Making the measurement accurate does not add value, since a person in the room also feels temperature differently as the conditions vary. \$\endgroup\$ – Sean Houlihane May 20 '16 at 10:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.